ViewDiff: 3D-Consistent Image Generation with Text-to-Image Models | Read Paper on Bytez